Multilabel Classification Evaluation using Ontology Information
|
|
- Buddy Greer
- 6 years ago
- Views:
Transcription
1 Multilabel Classification Evaluation using Ontology Information Stefanie Nowak, Hanna Lukashevich Fraunhofer Institute for Digital Media Technology IDMT Ehrenbergstrasse 31, Ilmenau, Germany, Abstract. Multilabel classification using ontology information is an emerging research area that combines machine learning methods with knowledge models. The performance assessment of such classification systems poses new challenges. We propose an evaluation measure that considers the mapping of label sets to their groundtruth and allows for the incorporation of real world knowledge. A distance-based measure from the area of hierarchical unilabel classification evaluation is extended to the case of multilabel classification and enriched with additional ontology information. The evaluation measure considers structure information, relationships and the agreement between annotators. 1 Introduction Traditional classification approaches classify multimedia documents in one of several categories. Applied to automatic image annotation that means that a photo depicts e.g., either a landscape, a city or persons. In reality, it is difficult to judge to which category a multimedia document exclusively belongs to. Mostly the documents contain aspects of different categories and should be labelled with all relevant items. This task is performed in multilabel classification. Figure 1 shows a simple hierarchical organization of concepts that can be assigned to photos in a multilabel annotation scenario. The hierarchy allows to make assumptions about the assignment of concepts to documents. E.g., if a photo is classified to contain trees, it also contains plants. Further semantic knowledge is provided if the concepts are organized in an ontology. Then, next to the is-a relationship of the hierarchical organization of concepts, additionally other relationships between concepts determine possible label assignments. The ontology restricts e.g., that for a certain sub-node only one concept can be assigned at a time disjoint items) or that a special concept like portrait) postulates other concepts like persons or animals. This paper discusses several approaches to multilabel classification evaluation and highlights the information that is used to assess the quality. It proposes a hierarchical multiannotation evaluation measure that incorporates ontology knowledge and can be used in a benchmarking scenario. This work has been supported by grant No. 01MQ07017 of the German THESEUS program, founded by the Federal Ministry of Economics and Technology.
2 4/14 4/14 4/14 4/14 4/14 Ground Truth System 1 System 2 Disjoint concepts Relationship 2/14 Link Costs 2/14 2/14 haspersons 2/14 2/14 2/14 2/14 Fig. 1. Visualization of an ontology fragment for image annotation. The concepts are hierarchical structured and different types of relationships are exemplarily highlighted. 2 Related work Many researchers regard the results of a multilabel classification system similar to the unilabel classification approach. The prediction is evaluated for each concept in isolation. This allows to use the well-known evaluation measures like precision, recall, f-measure or accuracy. E.g., Fan et al. use the accuracy in [1] to determine the quality of the classifier for each concept. The opposite way is to start with the multimedia document and evaluate if all concepts are assigned correctly. Then instead of comparing a single predicted label to a single groundtruth label, one needs to compare two sets of labels. As a result, the predicted labels can be fully correct label sets are identical), fully wrong the intersection of the sets is empty), or partly correct the sets have common labels, but are not fully identical). The Accuracy that is often used as measure in traditional classification evaluation cannot be used to judge partial matches, because it only regards one instance as correctly classified if all associated labels were correctly predicted. Partial matches are e.g., considered by utilizing the macro-average and micro-average f-measures, proposed by Tague in [2]. Shen et al. [3] proposed an α-evaluation and multilabel class recall and precision. α-evaluation generates a score while taking the ground-truth, predicted labels, missed labels and false positive labels into account. Moreover, false positives and missed labels can be penalized differently as it is more suitable for the particular application. The parameter α introduces the so-called forgiveness rate as a trade-off between the fully correct and partly correct prediction. Sun et al. propose in [4] semanticbased misclassification costs. Each class of documents is represented by a feature vector of all documents belonging to a certain category. The cosine distance between feature vectors of two categories is used as similarity measure and defines the average category similarity.
3 A hierarchical organization of concepts enables a hierarchical evaluation. Different hierarchical measures for unilabel classification are summarized in [5]. Intuitively the concepts, that are located near in a hierarchy are more similar than the ones that are located far. The idea is to judge an annotation from the predictor that does not match exactly to the groundtruth by their distance in the hierarchy. The most important measures are the depth independent distance-based misclassification costs and the depth dependent distance-based misclassification costs. In the former case, the predicted concept is compared to the correct one and the number of edges of the shortest path in the hierarchy between both are counted. In the latter case, an additional weight is assigned to each edge in the hierarchy. So, misclassifications in deeper levels of the hierarchy have lower costs than at an upper level. An evaluation measure for hierarchical multiclassification evaluation, is proposed by Blockeel et al. [6]. They extrapolate distances between individual labels to distances between sets of labels by mapping the feature vectors of the sets into Euclidean space where the individual labels form the base vectors. In [7] a hierarchical loss function is proposed that considers classification into a hierarchy with multiple and partial paths. The first wrongly classified node is regarded as mistake and adds to the loss while sub-nodes after a mistake are not considered. Underlying is the assumption that for each classification a path from root to leaf or from root to an internal node is present. They compare their work to the zero-one loss and symmetric-difference loss. 3 Evaluation Measure 3.1 Requirements on the Evaluation Measure The proposed evaluation measure is utilized in a benchmark for multilabel annotation of photos. 1 The participant s task is to annotate photos with 53 concepts in a multilabel annotation scenario. A small ontology of the concepts is provided that may be used for training the classifiers. To objectively compare the approaches of the participants, different conditions have to be assured. First, the evaluation measure should consider partial matches and deliver an annotation score for each image. Second, it has to be assured that a system that annotates plants instead of trees is judged better than a system that annotates mountains see Fig. 1). This issue is considered in the depth dependent distance-based misclassification costs, introduced in Sec. 2 for the unilabel annotation task. In a multilabel annotation scenario, the challenge is how to map the predicted label set to the groundtruth set. Third, the groundtruthing of the concepts depicted in an image, is no easy task, also not for humans. Some concept assignments are not objective and last in long discussions between the annotators. Therefore a study about the agreement on concepts between different annotators was conducted. Altogether 10 annotators annotated the same 100 images on their own. The degree of agreement for each concept over all photos was computed and should serve as a probability if the groundtruthing 1
4 was correct. E.g., for the concept clouds in 96% the annotators agreed on annotating the concept. In case of Aesthetic Image the annotators agreed only at 75%. These empirically obtained values are used as weighting factors for the calculated costs in case of misclassification. So, the more subjective concepts are weighted less than the more objective ones. Fourth, the relationships defined in the ontology should be kept. There should be an option to penalize a system, if it simultaneously annotates concepts that are defined as disjoint or ignores preconditions for relationships. None of the described evaluation measures in Sec. 2 fulfils these requirements. E.g., the hierarchical loss measure assumes that every node of the hierarchy possibly has annotation instances and that a continuous path exists through the hierarchy. In the ontology for image annotation, abstract nodes are defined that represent real-world knowledge but no visual concept. An example is the node representation. A portrait is a visual subconcept of representation and can be annotated by a classification system, but the concept representation itself is no visual concept. 3.2 Evaluation Measure for Multiannotation Scenarios Let us consider the predicted set of labels as P and the groundtruth set of labels as G. Each set contains labels l i respectively l j that are assigned to a multimedia document X. First, the false positive labels P = P \ P G) and the missed labels G = G \ P G) are computed. Please note that P + G P G is always valid, because the number of false positive and missed labels can never be greater than the number of the union of labels in both sets. Next, for each label l i from P a match to a label l j from G is calculated and for each label l j from G a mapping to a label l i from P is performed in an optimization procedure see Eq. 1). The costs between two labels l i and l j depend on the shortest path in the hierarchy between both concepts. Each link is associated with a cost that is cut in halves for each deeper level of the tree and is maximal equal to 1 for a path between two leaf nodes of the deepest level see Fig. 1). The costs c i for a 2 link are calculated as c i = i 1) with N as the number of links from the 2 N i=1 2i 1) deepest node to the root. If P =, the matching costs for all labels l j of G = G are set to the maximum. The matching costs are computed as follows: matchp, G) = l i P ) min costli, lj)) l j G alj ) + l j G ) min costli, lj)) alj) l i P 1) with l j = argmin lj G costli, lj)). al j ) determines the annotation agreement factor for a concept l j and ranges between [0, 1]. It is empirically determined through the agreement of different annotators on a concept, as described in 3.1. Optionally, a crosscheck on the predicted label set P is performed during computing matchp, G). If labels in P violate relationships from the ontology, these labels get the maximum costs of 1 as penalty assigned instead of calculating the minimal costs to a label of
5 G. Referring to the example in Fig. 1, a penalty of 1 is assigned if the system simultaneously annotates single and small group or if portrait is annotated without annotating one of the person concepts. Assuming that single is correct in the first example, the costs between small group from P and single from G are equal 1 instead of only equal 2/14 asingle) because of the hierarchy distances. scorex) = 1 matchp, G) P G ) α 2) The final score for each multimedia document X is based on the matching costs between P and G divided by the number of different concepts in both label sets see Eq. 2). The score is 1 if all concepts are correctly annotated and goes to 0 if no concept was found. Additionally, Shens α-factor, α 0), introduced in Sec. 2, is incorporated to weight the strictness of the score regarding fully and partly correct annotations, depending on the application demands. 4 Conclusion and Future Work In this paper, we propose an evaluation measure for the performance assessment of multiannotation classification systems incorporating ontology knowledge. A distance-based misclassification cost was extended from the unilabel to the multilabel case and further enriched with ontology information like its hierarchy, an annotation agreement factor and penalties for ignoring relationships. Next, an extensive evaluation of the behaviour of the measure in a real benchmarking scenario will be conducted. In future work, we would like to investigate how relationships in ontologies can be incorporated more differentiated in dependence from the evaluation scenario. Another point is how to base the evaluation measure on semantic similarity of concepts instead of distances in hierarchies, as the structure of the ontology is rather subjective and may change during time. References 1. Fan, J., Gao, Y., Luo, H., Jain, R.: Mining multilevel image semantics via hierarchical classification. IEEE Trans. on Multimedia 102) 2008) Tague, J.M.: Information retrieval experiment. In Jones, K.S., ed.: The pragmatics of information retrieval experimentation, Butterworths, London. 1981) Shen, X., Boutell, M., Luo, J., Brown, C.: Multi-label machine learning and its application to semantic scene classification. In: Intern. Symp. on Elec. Imag. 2004) 4. Sun, A., Lim, E.: Hierarchical text classification and evaluation. In: Proc. of the IEEE Intern. Conf. on Data Mining. Volume 528., California, USA 2001) 5. Freitas, A., de Carvalho, A.: A tutorial on hierarchical classification with applications in bioinformatics. Intelligent Information Technologies: Concepts, Methodologies, Tools and Applications 2007) Blockeel, H., Bruynooghe, M., Džeroski, S., Ramon, J., Struyf, J.: Hierarchical multi-classification. In: Proc. of Workshop on Multi-Relational Data Mining. 2002) 7. Cesa-Bianchi, N., Gentile, C., Zaniboni, L.: Incremental algorithms for hierarchical classification. The Journal of Machine Learning Research ) 31 54
Multimedia Information Retrieval
Multimedia Information Retrieval Prof Stefan Rüger Multimedia and Information Systems Knowledge Media Institute The Open University http://kmi.open.ac.uk/mmis Multimedia Information Retrieval 1. What are
More informationA Review of Performance Evaluation Measures for Hierarchical Classifiers
A Review of Performance Evaluation Measures for Hierarchical Classifiers Eduardo P. Costa Depto. Ciências de Computação ICMC/USP - São Carlos Caixa Postal 668, 13560-970, São Carlos-SP, Brazil Ana C. Lorena
More informationA Consumer Photo Tagging Ontology
A Consumer Photo Tagging Ontology Concepts and Annotations Stefanie Nowak Semantic Audiovisual Systems Fraunhofer IDMT Corfu, 29.09.2009 stefanie.nowak@idmt.fraunhofer.de slide 1 Outline Introduction Related
More informationAn Empirical Study of Lazy Multilabel Classification Algorithms
An Empirical Study of Lazy Multilabel Classification Algorithms E. Spyromitros and G. Tsoumakas and I. Vlahavas Department of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
More informationHIERARCHICAL MULTILABEL CLASSIFICATION
HIERARCHICAL MULTILABEL CLASSIFICATION Pooja D. Bankar 1 S.S.Sane 2 1 M.E. Student 2 HOD & Vice Principal 1,2 Department of Computer Engineering 1,2 K. K. W. I. E. E. R., Nashik Savitribai Phule Pune University,
More informationHow Reliable are Annotations via Crowdsourcing?
How Reliable are Annotations via Crowdsourcing? A Study about Inter-annotator Agreement for Multi-label Image Annotation ABSTRACT Stefanie Nowak Fraunhofer IDMT Ehrenbergstr. 31 98693 Ilmenau, Germany
More informationAnnotated Suffix Trees for Text Clustering
Annotated Suffix Trees for Text Clustering Ekaterina Chernyak and Dmitry Ilvovsky National Research University Higher School of Economics Moscow, Russia echernyak,dilvovsky@hse.ru Abstract. In this paper
More informationThe Fraunhofer IDMT at ImageCLEF 2011 Photo Annotation Task
The Fraunhofer IDMT at ImageCLEF 2011 Photo Annotation Task Karolin Nagel, Stefanie Nowak, Uwe Kühhirt and Kay Wolter Fraunhofer Institute for Digital Media Technology (IDMT) Ehrenbergstr. 31, 98693 Ilmenau,
More informationDomain-specific Concept-based Information Retrieval System
Domain-specific Concept-based Information Retrieval System L. Shen 1, Y. K. Lim 1, H. T. Loh 2 1 Design Technology Institute Ltd, National University of Singapore, Singapore 2 Department of Mechanical
More informationOntology Matching with CIDER: Evaluation Report for the OAEI 2008
Ontology Matching with CIDER: Evaluation Report for the OAEI 2008 Jorge Gracia, Eduardo Mena IIS Department, University of Zaragoza, Spain {jogracia,emena}@unizar.es Abstract. Ontology matching, the task
More informationFlatten Hierarchies for Large-scale Hierarchical Text Categorization
Flatten Hierarchies for Large-scale Hierarchical Text Categorization Xiao-Lin Wang 1, Bao-Liang Lu 1,2 1 Center for Brain-Like Computing and Machine Intelligence Department of Computer Science and Engineering
More informationThe Encoding Complexity of Network Coding
The Encoding Complexity of Network Coding Michael Langberg Alexander Sprintson Jehoshua Bruck California Institute of Technology Email: mikel,spalex,bruck @caltech.edu Abstract In the multicast network
More informationClassification with Class Overlapping: A Systematic Study
Classification with Class Overlapping: A Systematic Study Haitao Xiong 1 Junjie Wu 1 Lu Liu 1 1 School of Economics and Management, Beihang University, Beijing 100191, China Abstract Class overlapping has
More informationCSE 158. Web Mining and Recommender Systems. Midterm recap
CSE 158 Web Mining and Recommender Systems Midterm recap Midterm on Wednesday! 5:10 pm 6:10 pm Closed book but I ll provide a similar level of basic info as in the last page of previous midterms CSE 158
More informationAnnotation Component in KiWi
Annotation Component in KiWi Marek Schmidt and Pavel Smrž Faculty of Information Technology Brno University of Technology Božetěchova 2, 612 66 Brno, Czech Republic E-mail: {ischmidt,smrz}@fit.vutbr.cz
More informationExtraction of Web Image Information: Semantic or Visual Cues?
Extraction of Web Image Information: Semantic or Visual Cues? Georgina Tryfou and Nicolas Tsapatsoulis Cyprus University of Technology, Department of Communication and Internet Studies, Limassol, Cyprus
More informationAS DIGITAL cameras become more affordable and widespread,
IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 17, NO. 3, MARCH 2008 407 Integrating Concept Ontology and Multitask Learning to Achieve More Effective Classifier Training for Multilevel Image Annotation Jianping
More informationAn Empirical Study on Lazy Multilabel Classification Algorithms
An Empirical Study on Lazy Multilabel Classification Algorithms Eleftherios Spyromitros, Grigorios Tsoumakas and Ioannis Vlahavas Machine Learning & Knowledge Discovery Group Department of Informatics
More informationEFFICIENT INTEGRATION OF SEMANTIC TECHNOLOGIES FOR PROFESSIONAL IMAGE ANNOTATION AND SEARCH
EFFICIENT INTEGRATION OF SEMANTIC TECHNOLOGIES FOR PROFESSIONAL IMAGE ANNOTATION AND SEARCH Andreas Walter FZI Forschungszentrum Informatik, Haid-und-Neu-Straße 10-14, 76131 Karlsruhe, Germany, awalter@fzi.de
More informationFinding Tiny Faces Supplementary Materials
Finding Tiny Faces Supplementary Materials Peiyun Hu, Deva Ramanan Robotics Institute Carnegie Mellon University {peiyunh,deva}@cs.cmu.edu 1. Error analysis Quantitative analysis We plot the distribution
More informationInformation Retrieval System Based on Context-aware in Internet of Things. Ma Junhong 1, a *
Information Retrieval System Based on Context-aware in Internet of Things Ma Junhong 1, a * 1 Xi an International University, Shaanxi, China, 710000 a sufeiya913@qq.com Keywords: Context-aware computing,
More informationDefinition, Detection, and Evaluation of Meeting Events in Airport Surveillance Videos
Definition, Detection, and Evaluation of Meeting Events in Airport Surveillance Videos Sung Chun Lee, Chang Huang, and Ram Nevatia University of Southern California, Los Angeles, CA 90089, USA sungchun@usc.edu,
More informationA Short SVM (Support Vector Machine) Tutorial
A Short SVM (Support Vector Machine) Tutorial j.p.lewis CGIT Lab / IMSC U. Southern California version 0.zz dec 004 This tutorial assumes you are familiar with linear algebra and equality-constrained optimization/lagrange
More informationSupplementary Materials for Learning to Parse Wireframes in Images of Man-Made Environments
Supplementary Materials for Learning to Parse Wireframes in Images of Man-Made Environments Kun Huang, Yifan Wang, Zihan Zhou 2, Tianjiao Ding, Shenghua Gao, and Yi Ma 3 ShanghaiTech University {huangkun,
More informationSCENARIO BASED ADAPTIVE PREPROCESSING FOR STREAM DATA USING SVM CLASSIFIER
SCENARIO BASED ADAPTIVE PREPROCESSING FOR STREAM DATA USING SVM CLASSIFIER P.Radhabai Mrs.M.Priya Packialatha Dr.G.Geetha PG Student Assistant Professor Professor Dept of Computer Science and Engg Dept
More informationTexton Clustering for Local Classification using Scene-Context Scale
Texton Clustering for Local Classification using Scene-Context Scale Yousun Kang Tokyo Polytechnic University Atsugi, Kanakawa, Japan 243-0297 Email: yskang@cs.t-kougei.ac.jp Sugimoto Akihiro National
More informationColor Image Segmentation
Color Image Segmentation Yining Deng, B. S. Manjunath and Hyundoo Shin* Department of Electrical and Computer Engineering University of California, Santa Barbara, CA 93106-9560 *Samsung Electronics Inc.
More informationAn Experiment in Visual Clustering Using Star Glyph Displays
An Experiment in Visual Clustering Using Star Glyph Displays by Hanna Kazhamiaka A Research Paper presented to the University of Waterloo in partial fulfillment of the requirements for the degree of Master
More informationA CONSISTENCY MAINTENANCE OF SHARED BOUNDARY AFTER POLYGON GENERALIZATION
CO-182 A CONSISTENCY MAINTENANCE OF SHARED BOUNDARY AFTER POLYGON GENERALIZATION AI T.(1), ZHANG W.(2) (1) Wuhan University, WUHAN CITY, CHINA ; (2) Zhongnan University of Economics and Law, WUHAN CITY,
More informationAccess rights and collaborative ontology integration for reuse across security domains
Access rights and collaborative ontology integration for reuse across security domains Martin Knechtel SAP AG, SAP Research CEC Dresden Chemnitzer Str. 48, 01187 Dresden, Germany martin.knechtel@sap.com
More informationCluster-based Instance Consolidation For Subsequent Matching
Jennifer Sleeman and Tim Finin, Cluster-based Instance Consolidation For Subsequent Matching, First International Workshop on Knowledge Extraction and Consolidation from Social Media, November 2012, Boston.
More informationUnsupervised Learning and Clustering
Unsupervised Learning and Clustering Selim Aksoy Department of Computer Engineering Bilkent University saksoy@cs.bilkent.edu.tr CS 551, Spring 2009 CS 551, Spring 2009 c 2009, Selim Aksoy (Bilkent University)
More informationCharacter Recognition
Character Recognition 5.1 INTRODUCTION Recognition is one of the important steps in image processing. There are different methods such as Histogram method, Hough transformation, Neural computing approaches
More informationApplying Supervised Learning
Applying Supervised Learning When to Consider Supervised Learning A supervised learning algorithm takes a known set of input data (the training set) and known responses to the data (output), and trains
More informationCreating Large-scale Training and Test Corpora for Extracting Structured Data from the Web
Creating Large-scale Training and Test Corpora for Extracting Structured Data from the Web Robert Meusel and Heiko Paulheim University of Mannheim, Germany Data and Web Science Group {robert,heiko}@informatik.uni-mannheim.de
More informationC-NBC: Neighborhood-Based Clustering with Constraints
C-NBC: Neighborhood-Based Clustering with Constraints Piotr Lasek Chair of Computer Science, University of Rzeszów ul. Prof. St. Pigonia 1, 35-310 Rzeszów, Poland lasek@ur.edu.pl Abstract. Clustering is
More informationPattern recognition (4)
Pattern recognition (4) 1 Things we have discussed until now Statistical pattern recognition Building simple classifiers Supervised classification Minimum distance classifier Bayesian classifier (1D and
More informationGuidelines for the application of Data Envelopment Analysis to assess evolving software
Short Paper Guidelines for the application of Data Envelopment Analysis to assess evolving software Alexander Chatzigeorgiou Department of Applied Informatics, University of Macedonia 546 Thessaloniki,
More informationContext Ontology Construction For Cricket Video
Context Ontology Construction For Cricket Video Dr. Sunitha Abburu Professor& Director, Department of Computer Applications Adhiyamaan College of Engineering, Hosur, pin-635109, Tamilnadu, India Abstract
More informationBone Age Classification Using the Discriminative Generalized Hough Transform
Bone Age Classification Using the Discriminative Generalized Hough Transform Markus Brunk 1, Heike Ruppertshofen 1,2, Sarah Schmidt 2,3, Peter Beyerlein 3, Hauke Schramm 1 1 Institute of Applied Computer
More informationSemi-Supervised Clustering with Partial Background Information
Semi-Supervised Clustering with Partial Background Information Jing Gao Pang-Ning Tan Haibin Cheng Abstract Incorporating background knowledge into unsupervised clustering algorithms has been the subject
More informationHybrid Feature Selection for Modeling Intrusion Detection Systems
Hybrid Feature Selection for Modeling Intrusion Detection Systems Srilatha Chebrolu, Ajith Abraham and Johnson P Thomas Department of Computer Science, Oklahoma State University, USA ajith.abraham@ieee.org,
More informationPROJECT PERIODIC REPORT
PROJECT PERIODIC REPORT Grant Agreement number: 257403 Project acronym: CUBIST Project title: Combining and Uniting Business Intelligence and Semantic Technologies Funding Scheme: STREP Date of latest
More informationMandatory Leaf Node Prediction in Hierarchical Multilabel Classification
Mandatory Leaf Node Prediction in Hierarchical Multilabel Classification Wei Bi James T. Kwok Department of Computer Science and Engineering Hong Kong University of Science and Technology Clear Water Bay,
More informationProcess Model Consistency Measurement
IOSR Journal of Computer Engineering (IOSRJCE) ISSN: 2278-0661, ISBN: 2278-8727Volume 7, Issue 6 (Nov. - Dec. 2012), PP 40-44 Process Model Consistency Measurement Sukanth Sistla CSE Department, JNTUniversity,
More informationReference Model Concept for Structuring and Representing Performance Indicators in Manufacturing
Reference Model Concept for Structuring and Representing Performance Indicators in Manufacturing Stefan Hesse 1, Bernhard Wolf 1, Martin Rosjat 1, Dražen Nadoveza 2, and George Pintzos 3 1 SAP AG, SAP
More informationChapter 8: Enhanced ER Model
Chapter 8: Enhanced ER Model Subclasses, Superclasses, and Inheritance Specialization and Generalization Constraints and Characteristics of Specialization and Generalization Hierarchies Modeling of UNION
More informationKnowledge Guided Hierarchical Multi-Label Classification Over Ticket Data
246 IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT, VOL. 14, NO. 2, JUNE 2017 Knowledge Guided Hierarchical Multi-Label Classification Over Ticket Data Chunqiu Zeng, Wubai Zhou, Tao Li, Larisa Shwartz,
More informationarxiv: v1 [cs.mm] 12 Jan 2016
Learning Subclass Representations for Visually-varied Image Classification Xinchao Li, Peng Xu, Yue Shi, Martha Larson, Alan Hanjalic Multimedia Information Retrieval Lab, Delft University of Technology
More informationClustering Web Documents using Hierarchical Method for Efficient Cluster Formation
Clustering Web Documents using Hierarchical Method for Efficient Cluster Formation I.Ceema *1, M.Kavitha *2, G.Renukadevi *3, G.sripriya *4, S. RajeshKumar #5 * Assistant Professor, Bon Secourse College
More informationDrywall state detection in image data for automatic indoor progress monitoring C. Kropp, C. Koch and M. König
Drywall state detection in image data for automatic indoor progress monitoring C. Kropp, C. Koch and M. König Chair for Computing in Engineering, Department of Civil and Environmental Engineering, Ruhr-Universität
More informationR07. FirstRanker. 7. a) What is text mining? Describe about basic measures for text retrieval. b) Briefly describe document cluster analysis.
www..com www..com Set No.1 1. a) What is data mining? Briefly explain the Knowledge discovery process. b) Explain the three-tier data warehouse architecture. 2. a) With an example, describe any two schema
More informationMulti-Stage Rocchio Classification for Large-scale Multilabeled
Multi-Stage Rocchio Classification for Large-scale Multilabeled Text data Dong-Hyun Lee Nangman Computing, 117D Garden five Tools, Munjeong-dong Songpa-gu, Seoul, Korea dhlee347@gmail.com Abstract. Large-scale
More informationAAUITEC at ImageCLEF 2015: Compound Figure Separation
AAUITEC at ImageCLEF 2015: Compound Figure Separation Mario Taschwer 1 and Oge Marques 2 1 ITEC, Klagenfurt University (AAU), Austria, mario.taschwer@aau.at 2 Florida Atlantic University (FAU), Boca Raton,
More informationImproving Suffix Tree Clustering Algorithm for Web Documents
International Conference on Logistics Engineering, Management and Computer Science (LEMCS 2015) Improving Suffix Tree Clustering Algorithm for Web Documents Yan Zhuang Computer Center East China Normal
More informationA Connection between Network Coding and. Convolutional Codes
A Connection between Network Coding and 1 Convolutional Codes Christina Fragouli, Emina Soljanin christina.fragouli@epfl.ch, emina@lucent.com Abstract The min-cut, max-flow theorem states that a source
More informationLecture 18 Restoring Invariants
Lecture 18 Restoring Invariants 15-122: Principles of Imperative Computation (Spring 2018) Frank Pfenning In this lecture we will implement heaps and operations on them. The theme of this lecture is reasoning
More informationDecomposition of the output space in multi-label classification using feature ranking
Decomposition of the output space in multi-label classification using feature ranking Stevanche Nikoloski 2,3, Dragi Kocev 1,2, and Sašo Džeroski 1,2 1 Department of Knowledge Technologies, Jožef Stefan
More informationDeakin Research Online
Deakin Research Online This is the published version: Nasierding, Gulisong, Tsoumakas, Grigorios and Kouzani, Abbas Z. 2009, Clustering based multi-label classification for image annotation and retrieval,
More informationAn Intelligent Clustering Algorithm for High Dimensional and Highly Overlapped Photo-Thermal Infrared Imaging Data
An Intelligent Clustering Algorithm for High Dimensional and Highly Overlapped Photo-Thermal Infrared Imaging Data Nian Zhang and Lara Thompson Department of Electrical and Computer Engineering, University
More informationAutomatic Segmentation of Semantic Classes in Raster Map Images
Automatic Segmentation of Semantic Classes in Raster Map Images Thomas C. Henderson, Trevor Linton, Sergey Potupchik and Andrei Ostanin School of Computing, University of Utah, Salt Lake City, UT 84112
More informationData Mining with Oracle 10g using Clustering and Classification Algorithms Nhamo Mdzingwa September 25, 2005
Data Mining with Oracle 10g using Clustering and Classification Algorithms Nhamo Mdzingwa September 25, 2005 Abstract Deciding on which algorithm to use, in terms of which is the most effective and accurate
More informationMulticlass Classification
Multiclass Classification Instructor: Jessica Wu Harvey Mudd College The instructor gratefully acknowledges Eric Eaton (UPenn), David Kauchak (Pomona), Tommi Jaakola (MIT) and the many others who made
More informationImage Classification Using Wavelet Coefficients in Low-pass Bands
Proceedings of International Joint Conference on Neural Networks, Orlando, Florida, USA, August -7, 007 Image Classification Using Wavelet Coefficients in Low-pass Bands Weibao Zou, Member, IEEE, and Yan
More informationFuzzy Knowledge-based Image Annotation Refinement
284 Int'l Conf. IP, Comp. Vision, and Pattern Recognition IPCV'15 Fuzzy Knowledge-based Image Annotation Refinement M. Ivašić-Kos 1, M. Pobar 1 and S. Ribarić 2 1 Department of Informatics, University
More informationStochastic propositionalization of relational data using aggregates
Stochastic propositionalization of relational data using aggregates Valentin Gjorgjioski and Sašo Dzeroski Jožef Stefan Institute Abstract. The fact that data is already stored in relational databases
More informationA Text-Based Approach to the ImageCLEF 2010 Photo Annotation Task
A Text-Based Approach to the ImageCLEF 2010 Photo Annotation Task Wei Li, Jinming Min, Gareth J. F. Jones Center for Next Generation Localisation School of Computing, Dublin City University Dublin 9, Ireland
More informationCollection Guiding: Multimedia Collection Browsing and Visualization. Outline. Context. Searching for data
Collection Guiding: Multimedia Collection Browsing and Visualization Stéphane Marchand-Maillet Viper CVML University of Geneva marchand@cui.unige.ch http://viper.unige.ch Outline Multimedia data context
More informationChallenges on Combining Open Web and Dataset Evaluation Results: The Case of the Contextual Suggestion Track
Challenges on Combining Open Web and Dataset Evaluation Results: The Case of the Contextual Suggestion Track Alejandro Bellogín 1,2, Thaer Samar 1, Arjen P. de Vries 1, and Alan Said 1 1 Centrum Wiskunde
More informationDIGITAL VIDEO now plays an increasingly pervasive role
IEEE TRANSACTIONS ON MULTIMEDIA, VOL. 9, NO. 5, AUGUST 2007 939 Incorporating Concept Ontology for Hierarchical Video Classification, Annotation, and Visualization Jianping Fan, Hangzai Luo, Yuli Gao,
More informationSupplementary Note to Detecting Building-level Changes of a City Using Street Images and a 2D City Map
Supplementary Note to Detecting Building-level Changes of a City Using Street Images and a 2D City Map Daiki Tetsuka Takayuki Okatani Graduate School of Information Sciences, Tohoku University okatani@vision.is.tohoku.ac.jp
More informationImage retrieval based on bag of images
University of Wollongong Research Online Faculty of Informatics - Papers (Archive) Faculty of Engineering and Information Sciences 2009 Image retrieval based on bag of images Jun Zhang University of Wollongong
More informationCSE Data Mining Concepts and Techniques STATISTICAL METHODS (REGRESSION) Professor- Anita Wasilewska. Team 13
CSE 634 - Data Mining Concepts and Techniques STATISTICAL METHODS Professor- Anita Wasilewska (REGRESSION) Team 13 Contents Linear Regression Logistic Regression Bias and Variance in Regression Model Fit
More informationRecognizing Apples by Piecing Together the Segmentation Puzzle
Recognizing Apples by Piecing Together the Segmentation Puzzle Kyle Wilshusen 1 and Stephen Nuske 2 Abstract This paper presents a system that can provide yield estimates in apple orchards. This is done
More informationMidterm Examination CS540-2: Introduction to Artificial Intelligence
Midterm Examination CS540-2: Introduction to Artificial Intelligence March 15, 2018 LAST NAME: FIRST NAME: Problem Score Max Score 1 12 2 13 3 9 4 11 5 8 6 13 7 9 8 16 9 9 Total 100 Question 1. [12] Search
More informationConstruct an optimal tree of one level
Economics 1660: Big Data PS 3: Trees Prof. Daniel Björkegren Poisonous Mushrooms, Continued A foodie friend wants to cook a dish with fresh collected mushrooms. However, he knows that some wild mushrooms
More informationQuery Difficulty Prediction for Contextual Image Retrieval
Query Difficulty Prediction for Contextual Image Retrieval Xing Xing 1, Yi Zhang 1, and Mei Han 2 1 School of Engineering, UC Santa Cruz, Santa Cruz, CA 95064 2 Google Inc., Mountain View, CA 94043 Abstract.
More informationA Comparison of Global and Local Probabilistic Approximations in Mining Data with Many Missing Attribute Values
A Comparison of Global and Local Probabilistic Approximations in Mining Data with Many Missing Attribute Values Patrick G. Clark Department of Electrical Eng. and Computer Sci. University of Kansas Lawrence,
More informationTRANSPARENT OBJECT DETECTION USING REGIONS WITH CONVOLUTIONAL NEURAL NETWORK
TRANSPARENT OBJECT DETECTION USING REGIONS WITH CONVOLUTIONAL NEURAL NETWORK 1 Po-Jen Lai ( 賴柏任 ), 2 Chiou-Shann Fuh ( 傅楸善 ) 1 Dept. of Electrical Engineering, National Taiwan University, Taiwan 2 Dept.
More informationA Hierarchical Document Clustering Approach with Frequent Itemsets
A Hierarchical Document Clustering Approach with Frequent Itemsets Cheng-Jhe Lee, Chiun-Chieh Hsu, and Da-Ren Chen Abstract In order to effectively retrieve required information from the large amount of
More informationApplication Networking Trends
HIGHLIGHTS OF ASN SURVEY RESULTS n The Futuriom Applications Specific Networking (ASN) Survey collected detailed feedback from 2 enterprise technology end users on the topic of networking cloud applications.
More informationCS473: Course Review CS-473. Luo Si Department of Computer Science Purdue University
CS473: CS-473 Course Review Luo Si Department of Computer Science Purdue University Basic Concepts of IR: Outline Basic Concepts of Information Retrieval: Task definition of Ad-hoc IR Terminologies and
More informationIJREAT International Journal of Research in Engineering & Advanced Technology, Volume 1, Issue 5, Oct-Nov, 2013 ISSN:
Semi Automatic Annotation Exploitation Similarity of Pics in i Personal Photo Albums P. Subashree Kasi Thangam 1 and R. Rosy Angel 2 1 Assistant Professor, Department of Computer Science Engineering College,
More informationWhich Role for an Ontology of Uncertainty?
Which Role for an Ontology of Uncertainty? Paolo Ceravolo, Ernesto Damiani, Marcello Leida Dipartimento di Tecnologie dell Informazione - Università degli studi di Milano via Bramante, 65-26013 Crema (CR),
More informationAutomated Visualization Support for Linked Research Data
Automated Visualization Support for Linked Research Data Belgin Mutlu 1, Patrick Hoefler 1, Vedran Sabol 1, Gerwald Tschinkel 1, and Michael Granitzer 2 1 Know-Center, Graz, Austria 2 University of Passau,
More informationDistributed Shared Memory Consistency Object-based Model
Journal of Computer Science 3 (1): 57-61, 27 ISSN1549-3636 27 Science Publications Corresponding Author: Distributed Shared Memory Consistency Object-based Model Abdelfatah Aref Yahya and Rana Mohamad
More informationSemantic text features from small world graphs
Semantic text features from small world graphs Jurij Leskovec 1 and John Shawe-Taylor 2 1 Carnegie Mellon University, USA. Jozef Stefan Institute, Slovenia. jure@cs.cmu.edu 2 University of Southampton,UK
More informationA Bagging Method using Decision Trees in the Role of Base Classifiers
A Bagging Method using Decision Trees in the Role of Base Classifiers Kristína Machová 1, František Barčák 2, Peter Bednár 3 1 Department of Cybernetics and Artificial Intelligence, Technical University,
More informationGeneric object recognition using graph embedding into a vector space
American Journal of Software Engineering and Applications 2013 ; 2(1) : 13-18 Published online February 20, 2013 (http://www.sciencepublishinggroup.com/j/ajsea) doi: 10.11648/j. ajsea.20130201.13 Generic
More informationA Multiclassifier based Approach for Word Sense Disambiguation using Singular Value Decomposition
A Multiclassifier based Approach for Word Sense Disambiguation using Singular Value Decomposition Ana Zelaia, Olatz Arregi and Basilio Sierra Computer Science Faculty University of the Basque Country ana.zelaia@ehu.es
More informationMobile Cloud Multimedia Services Using Enhance Blind Online Scheduling Algorithm
Mobile Cloud Multimedia Services Using Enhance Blind Online Scheduling Algorithm Saiyad Sharik Kaji Prof.M.B.Chandak WCOEM, Nagpur RBCOE. Nagpur Department of Computer Science, Nagpur University, Nagpur-441111
More informationAS STORAGE and bandwidth capacities increase, digital
974 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 13, NO. 7, JULY 2004 Concept-Oriented Indexing of Video Databases: Toward Semantic Sensitive Retrieval and Browsing Jianping Fan, Hangzai Luo, and Ahmed
More informationContextual Classification with Functional Max-Margin Markov Networks
Contextual Classification with Functional Max-Margin Markov Networks Dan Munoz Nicolas Vandapel Drew Bagnell Martial Hebert Geometry Estimation (Hoiem et al.) Sky Problem 3-D Point Cloud Classification
More informationA Novel Approach for Minimum Spanning Tree Based Clustering Algorithm
IJCSES International Journal of Computer Sciences and Engineering Systems, Vol. 5, No. 2, April 2011 CSES International 2011 ISSN 0973-4406 A Novel Approach for Minimum Spanning Tree Based Clustering Algorithm
More informationA Multiclassifier based Approach for Word Sense Disambiguation using Singular Value Decomposition
A Multiclassifier based Approach for Word Sense Disambiguation using Singular Value Decomposition Ana Zelaia, Olatz Arregi and Basilio Sierra Computer Science Faculty University of the Basque Country ana.zelaia@ehu.es
More informationAutomatic image annotation refinement using fuzzy inference algorithms
16th World Congress of the International Fuzzy Systems Association (IFSA) 9th Conference of the European Society for Fuzzy Logic and Technology (EUSFLAT) Automatic image annotation refinement using fuzzy
More informationInteractive Fault Localization Using Test Information
Hao D, Zhang L, Xie T et al. Interactive fault localization using test information. JOURNAL OF COMPUTER SCIENCE AND TECHNOLOGY 24(5): 962 974 Sept. 2009 Interactive Fault Localization Using Test Information
More informationWATERSHED, HIERARCHICAL SEGMENTATION AND WATERFALL ALGORITHM
WATERSHED, HIERARCHICAL SEGMENTATION AND WATERFALL ALGORITHM Serge Beucher Ecole des Mines de Paris, Centre de Morphologie Math«ematique, 35, rue Saint-Honor«e, F 77305 Fontainebleau Cedex, France Abstract
More informationSemi supervised clustering for Text Clustering
Semi supervised clustering for Text Clustering N.Saranya 1 Assistant Professor, Department of Computer Science and Engineering, Sri Eshwar College of Engineering, Coimbatore 1 ABSTRACT: Based on clustering
More informationRetrieval Evaluation. Hongning Wang
Retrieval Evaluation Hongning Wang CS@UVa What we have learned so far Indexed corpus Crawler Ranking procedure Research attention Doc Analyzer Doc Rep (Index) Query Rep Feedback (Query) Evaluation User
More information